Are any computer scientists here on Mastodon working in Visual Computing, or Computer Vision? I'm looking for a collaborator to work on the visualisation of medieval handwriting as movement. Would be grateful for re-posts!
@litteracarolina an assertive question to explore further context: do I understand correctly that you're working on deriving writing gestures (mostly) from the scriptures? Sounds intriguing, but if my understanding is correct, the computing subfield might be somewhat different. E.g. computer animation, reverse kinematics? Not sure
@litteracarolina @mapto it's an inverse problem, isn't it? A drawing program might have a model of a pen that has a certain shaped tip that moves through a path specified via a digitizer tablet (e.g. line gets thicker or thinner depending on how you hold it.)
Sounds like you have the image and you want to find a path that creates that image that is the "most likely" from some probability distribution.
@UP8 @mapto That's roughly it, but I don't want to re-create images. I want to extract the paths of pen movements from historical writing & analyse them en masse to see what patterns & differences emerge from scripts & scribes. This is a multi-step problem, with something like a neural network for extraction, followed by comparison using various techniques. I don't know what techniques exactly, & advice would be great, but I need a collaborator in any case if I want to write a funding proposal.
@litteracarolina @mapto but it sounds like you have the image and you want to get something like https://www.w3.org/TR/InkML/ or https://developer.wacom.com/en-us/products/universal-ink-model ?
@UP8 These are both really helpful tools to know about, thank you! But as far as I understand, they work via digital pens. Historic writing exists as pixel-based images of manuscripts without additional embedded data. Tracing over the writing with a digital pen would take too long. It's also not enough to have just the coordinates; there is a directionality to the movement that is key for historic writing.
@litteracarolina but would you want data in that format if you could get it by software looking at the image without anybody tracing anything? if not, what exact information do you need?
@UP8 Yes, potentially. I'm interested in figuring out how the underlying movement of the hand/pen confers individuality on a person's handwriting. This might include information about stroke speed, direction and length, ink dipping rates, pen angle based on stroke width, etc. But there might be better ways of analysing movement that I don't know about.
@mapto Thank you, this is a very good question! It's not just about visualisation, but also about working out efficient ways of predicting and analysing the movement of the pen during writing. I'm also unsure of the field that this would belong in. Animation doesn't feel right, but you may be right that something like kinematics could also be applicable.